Research ArticleEvaluating the Acute Effect of Stereoscopic Recovery by DichopticStimulation Using Electroencephalogram

Wei Shi ,1 Luyang He ,2 Bin Lv ,2 Li Li ,1 and Tongning Wu 2

1Department of Ophthalmology, Beijing Children’s Hospital, Capital Medical University, Beijing, China2China Academy of Information and Communications Technology, Beijing, China

Correspondence should be addressed to Tongning Wu; wutongning@caict.ac.cn

Received 29 February 2020; Revised 19 March 2020; Accepted 23 March 2020; Published 13 April 2020

Guest Editor: Chenxi Huang

Copyright © 2020 Wei Shi et al. This is an open access article distributed under the Creative Commons Attribution License, whichpermits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Amblyopia is a common developmental disorder in adolescents and children. Stereoscopic loss is a symptom of amblyopia that canseriously affect the quality of patient’s life. Recent studies have shown that the push-pull perceptual learning protocol had a positiveeffect on stereoscopic recovery. In this study, we developed a stereoscopic training method using a polarized visualization systemaccording to the push-pull protocol. Dichoptic stimulation for 36 anisometropic and amblyopic subjects and 33 children withnormal visual acuity (VA) has been conducted. Electroencephalogram (EEG) was used to evaluate the neurophysiologicalchanges before, during, and after stimulation. For the anisometropic and amblyopic subjects, the statistical analysisdemonstrated significant differences (p < 0:01) in the beta rhythm at the middle temporal and occipital lobes, while the EEGfrom the normal VA subjects indicated no significant changes when comparing the results before and after training. Weconcluded that the dichoptic training in our study can activate the middle temporal visual area and visual cortex. The EEGchanges can be used to evaluate the training effects. This study also found that the beta band EEG acquired during visualstimulation at the dorsal visual stream can be potentially used for predicting acute training effect. The results facilitated theoptimization of the individual training plan.

1. Introduction

Amblyopia is a common developmental visual disorder [1],which reduces the visual acuity (VA) of one or both eyeswithout obvious defects in the visual pathway, and cannotbe solved immediately using eyeglasses or contact lenses[2]. Patients with amblyopia usually present several visualfunctional deficiencies, such as refractive errors, low sensitiv-ity to contrast or dynamic objects, and limited stereopsis [3,4], which is estimated to affect 1-5% of the adult populationand even more for children [2, 4]. Stereoscopic vision isone of the most advanced visual functions, which providesa sense of depth in the environment and helps develop basicskills, such as grasping, catching, and walking around obsta-cles at a high speed [5]. Hence, poor stereoscopic depth per-ception can seriously affect the quality of patient’s life.Therefore, it has clinical importance to develop treatmentmethods for the recovery of stereopsis.

The conventional methods for stereoscopic recoveryincluded monocular occlusion therapy and therapeutic drugssuch as atropine [2, 6, 7]. These methods cover or blur thesound eye to force the use of the amblyopic one [8], withthe hypothesis that shielding the function of the strong eyefacilitated the development of the weak eye. As consequence,the disparity of VA would be levelled. These treatments areoften exclusive to children, and adults with stereoscopic def-icits are considered untreatable [2]. The use of monocles hasbeen reported to help 50% to 85% of amblyopia childrenachieve normal VA [6], but with a significant recurrence rateas high as 27% [9]. Dichoptic treatment is another effectiveapproach to solve stereoscopic vision problems [2]. Thistreatment presents different images to both eyes simulta-neously, aiming to reduce intraocular suppression, which isthe primary cause for multiple visual deficits [10]. Basedon this concept, a series of dichoptic perceptual learningmethods became available, including dichoptic videos [11,

HindawiComputational and Mathematical Methods in MedicineVolume 2020, Article ID 9497369, 7 pageshttps://doi.org/10.1155/2020/9497369

12], three-dimensional (3D) video games [13, 14], and virtualreality systems [15, 16]. These studies suggested that VA andstereo sensitivity can be improved through 5 to 30 hours ofperceptual learning sessions for children and adults [14,15]. Among the dichoptic perceptual learning methods, Ooiet al. [17] proposed a push-pull perceptual learning protocol.This protocol can effectively reduce sensory eye dominanceand enhance binocular balance [17, 18]. It was designed tostimulate the weak eye (push), while suppressing the percep-tion of the strong eye (pull) in order to recalibrate the binoc-ular balance of excitatory and inhibitory interactions. Incontrast, the conventional push-only protocols solely stimu-lated the weak eye without inhibiting the strong one. Thistreatment shifted the balance between two eyes towards theweak one and thus improved stereopsis. Numerous studiesindicated that this protocol would be promising for stereo-scopic recovery [2, 18, 19].

The widely used methods for evaluating stereoscopicrecovery were based on clinical examinations, e.g., synopto-phore examination, Randot circles test, and Titmus test[20]. These tests focused on assessing stereo sensitivitythrough behavioural reactions, which may depend on subjec-tive description [2]. Moving toward an objective evaluationfor stereoscopic recovery, recent studies have implementedneurophysiologic methods such as functional magnetic reso-nance imaging (fMRI) [21–25] and functional near-infraredspectroscopy (fNIRS) [26]. The results demonstrated thecapability in evaluating stereoscopic recovery. However,those methods required expensive equipment and inconve-nient operations, especially unsuitable for children. Besides,the abovementioned methods were incapable of monitoringreal-time neurophysiological changes during training. Tosolve the problem, electroencephalogram (EEG) has alsobeen used in the relevant study. For example, Deng et al.assessed binocular processing differences following percep-tual learning in adult anisometropic patients using steady-state visual evoked potential [27].

In our study, we developed a dichoptic stimulation sys-tem based on the push-pull protocol. The applicability ofEEG to assess the training effect was investigated. The possi-bility of using EEG as a biomarker to predict the acute train-ing effect has been discussed as well.

2. Materials and Methods

2.1. Subjects. In the present study, 69 children (4 to 17 yearsold, mean ± standard deviation: 7:1 ± 2:7 years old) wererecruited from the Department of Ophthalmology of BeijingChildren’s Hospital, Capital Medical University from April2017 to August 2018. Thirty-six children were diagnosed

with anisometropia and amblyopia, and the other 33 werechildren with normal VA. All subjects and their parentswere informed about the experimental protocol, and writ-ten informed consent was obtained. Our experiments wereapproved by the ethics committee of Beijing Children’sHospital.

2.2. EEG Data Acquisition and Associated Procedures. EEGdata were recorded using a NeuroScan SynAmps2 64-channel EEG amplifier (Compumedics, Victoria, Australia)with a 64-channel elastic cap (Quik-Cap, NeuroScan) at asampling rate of 1000Hz. The impedance of all electrodeswas adjusted to less than 10 kΩ, and the mean values of M1and M2 (according to the international 10-20 EEG system)were set as the reference. Before training, all subjects werewell rested and stayed alert while sitting in a comfortablechair. At the beginning of training, subjects were told to lookat a blank wall, and EEG signals were recorded for 5 minutes.During the stimulation, subjects placed their head in a chinr-est to ensure screen alignment (the centre of the two eyes wasaligned to the centre of the screen) and observed the stimulifor 20 minutes. EEG signals were recorded for 20 minutessimultaneously. We then recorded the EEG signal for 5minutes after the visual stimulus, while the subjects lookedat the blank wall again. The experimental flow chart is shownin Figure 1.

2.3. Dichoptic Stimulation. The dichoptic stimulation wasperformed in a clinical assessment room with constant lumi-nance. The experiment was conducted using a PC and an LGD2343P polarized 3D monitor (LG Electronics, Seoul, SouthKorea) with a resolution of 1920 × 1080 and a refreshing rateof 120Hz. All subjects wore polarized glasses and their cor-rective lenses. The visual stimuli were developed using thepush-pull perceptual learning protocol and presented at anobservation distance of 80 cm, with a background brightnessof 35 candelas per square meter (cd/m2). During the stimu-lation, the rivalling half image to the amblyopic eye was per-ceived (push), while the half image to the strong eye wasperceptually suppressed (pull) (see Figure 2). The image sizefor the strong eye was 200 × 200 pixels, and the image sizefor the weak eye had four levels: 200 × 200, 400 × 400, 600× 600, and 800 × 800 pixels. The contrast ratio for weakeye was set as 100% during the stimulation, while 5% to50% for the strong eye. The selection of the image size andcontrast ratio depended on the extent of the VA disparity.For normal VA subjects, the image size for both eyes wasset at 200 × 200 pixels, and contrast ratio was set at 100%.Figure 3 shows an experimental scenario in which a subject

20 mins

Polarized screen

During stimulation

Break

Before stimulation

5 mins

Blank wall

5 mins

After stimulation

Blank wall

1 min

Break

1 min

Figure 1: Experimental flowchart. This experiment procedure consisted of three steps: before, during, and after stimulation. In each step, thesubjects were told to look at blank wall or polarized screen while recording EEG. Between each step, there is a 1-minute break.

2 Computational and Mathematical Methods in Medicine

was looking at the stimulus image on the screen, while theEEG was being recorded.

2.4. EEG Data Analysis and Statistical Analysis. The raw EEGsignals were processed offline using Curry 7.0 (Compu-medics, Victoria, Australia) and MATLAB (MathWorks,Natick, MA). The preprocessing of the raw EEG dataincluded the use of band-pass filters (0.1-30Hz), rereferen-

cing (the mastoids are chosen as reference electrodes), andindependent component analysis (ICA) for artefact removal(including eye movements, temporal muscle activity, and lin-ear noise). The procedures were realised by EEGLAB [28].We calculated the power spectral density (PSD) of each elec-trode for three different states, i.e., before, during, and afterstimulation, using Welch method [29]. Moreover, four fre-quency bands were analyzed in the study, i.e., delta (0.1 to4Hz), theta (4 to 8Hz), alpha (8 to 14Hz), and beta (14 to30Hz). To reduce the impact of individual differences, theabsolute PSD in each frequency band was normalized bydividing by the total PSD to obtain the relative PSD, whichwas used for statistical analysis.

A two-way analysis of variance (ANOVA) was per-formed (per channel and per frequency band) to evaluatethe resulting PSD changes with the factors of VA disparityand stimulation state. VA was converted to the logarithmof the minimum angle of resolution (LogMAR). Accordingly,the subjects were divided into three levels: 0 to 0.2 (2 lines inthe visual chart of LogMAR) for mild VA disparity, 0.2 to 0.4(2 to 4 lines) for moderate VA disparity, and above 0.4 forsevere VA disparity (exceeding 4 lines). There were 10 sub-jects with mild VA disparity, 13 subjects with moderate VAdisparity, and 13 with severe VA disparity. The three

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

Training stimulus

Weak eye Strong eye Weak eye Strong eye

Figure 2: Push-pull training scenario used in the study. The initial images perceived by the two eyes are shown in (a). During the stimulationprocess, the images seen by both eyes were synchronously processed and switched every 500 milliseconds and looped from (a) to (j) over 5seconds, allowing the half image viewed by the weak amblyopic eye to be perceived and the one viewed by the strong eye to be suppressed.

Figure 3: Experimental scenario. One subject was looking at apolarized screen while recording EEG.

3Computational and Mathematical Methods in Medicine

stimulation states were before, during, and after stimulation.In order to assess the effect relating to different ages, we alsodivided 69 subjects into two groups, i.e., 16 patients plus 18normal VA subjects in the group below 7 years old and 20patients plus 15 normal VA subjects in the group exceeding7 years old. We performed independent t tests for the normalVA subjects and the patients, respectively, to assess thebetween-group difference.

Moreover, it would be of great interest to investigate thecorrelation between the EEG change during stimulationand the treatment effect, because it can provide insight onpredicting the individual recovery effect which enabled theoptimization of the training contents. The statistical experi-ments were designed for the 36 children with anisometropiaand amblyopia. Correlation between recovery effect (calcu-lated by subtracting the poststimulation VA disparity fromthe prestimulation VA disparity) and the change of relativePSD was investigated. The analyses were conducted perchannel and per band.

The statistical analyses were performed using SPSS 25.0(IBM; Armonk, NY).

3. Results

Figure 4 illustrates the EEG topographic maps averagedacross all 36 amblyopia subjects and 33 normal VA subjectsin four frequency bands. Two-way ANOVA yielded signifi-cant differences for the interaction effect of VA disparityand stimulation state in the beta pattern, TP7, p = 0:004,F ð4, 198Þ = 4:016. There were no other channels or fre-quency bands with significant differences for the interactioneffect. For the main effect of VA disparity, we found signifi-cant differences in the delta band (C6, T8, C1, CZ, CP1,CPZ, p < 0:01) and the theta band (F7, F5, F6, F8, TP7, T8,p < 0:01). For the main effect of stimulation state, significantdifferences were found in the beta (C1, C6, P1, P6, T7, P8,TP7, TP8, O1, O2, OZ, p < 0:01) and delta bands (FP1,FPZ, FP2, AF3, AF4, F7, F5, F6, F8, TP7, T8, p < 0:01). Weused the least significant difference (LSD) for the post hoctest. In terms of after vs. before stimulation, the post hoc testyielded significant differences in the beta bands across sixchannels: T7 (p = 0:009), P8 (p = 0:007), TP8 (p = 0:008),O1 (p = 0:004), OZ (p = 0:001), and O2 (p = 0:007). Therewas no significant change after vs. before stimulation inthe delta frequency bands. We also found significant differ-ences in the beta and delta rhythms after vs. during stim-ulation (beta: all channels; delta: FP1, FPZ, FP2, AF3,AF4, F7, F5, F6, p < 0:01), including occipital areas and thetemporal lobes.

To compare the EEG differences between normal sub-jects and subjects with anisometropia, we performed t testsfor before and after the stimulation on the relative PSDs ofthe four frequency bands of the 33 normal VA subjects andthe 36 anisometropic subjects (per channel). The t test resultsfrom the normal VA subjects revealed no significant differ-ences across all channels and all frequency bands, but theresults from the anisometropic patients revealed significantdifferences in the beta frequency PSDs on 25 channels(FT7, FC5, FC3, T7, C5, C3, C4, C6, T8, TP7, CP5, CP3,

CP4, CP6, TP8, P5, P3, P1, P4, P6, P8, O1, OZ, O2). Theanisometropic subjects had no significant differences in anyof the channels in the alpha, theta, or delta bands. The t testsfor between-group difference showed no significant changesfrom the normal VA subjects and the patients in the twoage groups.

The correlation analysis showed that high correlation(r > 0:8) for two channels at occipital region (O1 and OZ).Five channels at bilateral middle temporal regions (FT7, T8,T7, P8, and Tp8) showed correlation coefficients between0.5 and 0.8. The detected significant correlation was foundat beta band. The channels are shown in Figure 5.

4. Discussion

After visual training, amblyopic patients showed significantlychanged beta band PSD at occipital and middle temporalregions. In contrast, no difference was detected betweenbefore and after training from normal VA subjects. The find-ing revealed that the stimulation based on the push-pull pro-tocol had limited effects on normal VA subjects. Thecomparison showed no difference between the two agegroups, which may indicate that the applied stimulus patternwas not age-dependent.

Amblyopia is believed to result from a dysfunction of thecortical developmental processes [23, 24], although the exactextent of this visual cortical deficit is currently unknown. As aresult, functional neuroimaging techniques have been used tostudy the changes at the neurophysiological level after treat-ment [25]. For example, Yang et al. studied blood oxygenlevel-dependent activity in amblyopia patients before andafter levodopa treatment [24]. Similarly, Gupta et al. studiedhemodynamic changes in individuals with strabismic ambly-opia using fMRI and diffusion tensor imaging before andafter eye patching [25]. In addition, Iwata et al. used fNIRSto assess the differences in oxygenated haemoglobin concen-tration changes between binocular and monocular treatment[26]. Their results revealed that during or after stimulation,the occipital region demonstrated neurophysiological differ-ences, such as an elevated number of active voxels or bloodoxygen levels, resulting in improved stereopsis after stimula-tions. Numerous studies have shown that during visual stim-ulation, beta band or gamma band power increases inoccidental-parietal brain regions [27, 28]. The increase ofhigh-frequency power was associated with enhanced neuro-nal discharge activity [30], which was likely linked to theincrease of blood-oxygen concentration. The change in thedelta band may be associated with fatigue. In all, the EEGresults were consistent with reports using fNIRS and fMRI.In view of its mobility and convenience, EEG was advanta-geous in evaluating the acute effect of stereoscopic training.

There was a significant difference in beta power spectraldensity in the primary visual cortex, and there was also a sig-nificant difference in the middle temporal lobe region, whereMT was located [31]. MT was a specific region of the dorsalextrastriata processing stream, which was highly motion sen-sitive and can integrate local motion signals into a coherentmotion representation [32, 33]. In fact, converging evidencefrom human psychophysics and animal neurophysiology

4 Computational and Mathematical Methods in Medicine

has indicated that stereopsis was associated with abnormalMT function. Our study showed that after visual stimulation,the temporal lobe area of the anisometropic subjects was

active (compared to before stimulation), while that of thenormal subjects was not, suggesting that MT played animportant role in the formation of anisometropia. Dichoptic

Beta Alpha Theta Delta

Beforestimulation

Duringstimulation

Afterstimulation

0.1 0.15 0.2 0.25 0.3 0.08 0.09 0.1 0.11 0.13 0.14 0.14 0.16 0.18 0.2 0.22 0.45 0.5 0.55 0.60.12

(a)

Beta Alpha Theta Delta

Beforestimulation

Duringstimulation

Afterstimulation

0.1 0.15 0.2 0.25 0.3 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.14 0.16 0.18 0.2 0.22 0.5 0.54 0.58 0.62

(b)

Figure 4: Mean EEG topographic maps over 36 anisometropic subjects (a) and 33 normal VA subjects (b). Each column represents thedistribution of the averaged relative PSDs of the EEGs for the four frequency bands, while each row represents different stimulation states(before, during, and after stimulation).

5Computational and Mathematical Methods in Medicine

stimulation in our study was able to stimulate MT as well asthe visual cortex, thus possibly creating a higher chance ofrestoring the stereo sensitivity of anisometropia.

Analysis focusing on patients with anisometropia andamblyopia clearly demonstrated that the relative PSD changeduring training and the training effect correlated significantlyalong the dorsal extrastriata processing stream. The mostintense correlation was found at area V1, which was essentialto the conscious processing of visual stimuli [34]. In addition,damage to V1 may lead to disruption or even loss specificaspects of vision (e.g., depth perception [35]). The strong cor-relation between the PSD change and the posttraining effectalong the dorsal visual stream revealed that these regionsmay closely associate with the acute treatment effect. It alsoindicated the possibility to use the EEG data from this areaas an indicator for choosing the desired training contents.As such, it has the potential to act as a biomarker for predict-ing the stimulating effect and for adjusting the individualtraining plan (the stimulation length or the content). The fea-tures derived from the EEG on these channels may also beemployed as a classifier to identify the appropriate stimula-tion method for the patients, which would be investigatedour further studies when more subjects were recruited.

The limitation of the study was the limit number of thesubjects under evaluation. In addition, we did not track theirlong-term training effect due to the mobility of the subjects.

5. Conclusions

In our study, we focused on the stereoscopic recovery ofamblyopic patients. We designed a dichoptic stimulation sys-tem adopted from push-pull perceptual learning using polar-

ized glasses, performed training on 33 normal VA childrenand 36 anisometropic children, and evaluated its acuteeffects through EEG. After a 20-min dichoptic stimulation,we detected enhanced beta rhythm PSD after stimulationin the visual cortex and MT of the anisometropic subjects.Our study provided evidence indicating that dichoptictraining was able to stimulate MT and the primary visualcortex, and EEG acquired in those regions has potentialapplications in evaluating the acute effect of stereoscopictraining. The study also discussed the possibility of usingthe EEG signal as biomarker to predict the acute trainingeffect. In future studies, we will enroll more subjects andinvestigate the use of this EEG feature to optimizing theindividual training plan.

Data Availability

The EEG data used to support the findings of this study areavailable from the corresponding author upon request.

Conflicts of Interest

The authors declare that there is no conflict of interestregarding the publication of this paper.

Acknowledgments

This work was supported by the Beijing Scientific Researchand Development Program, [PX2017003], National Scienceand Technology Major Project, [2018ZX10301201], Wu Jiep-ing Medical Foundation, [320.6750.17179], and National Nat-ural Science Foundation Project, [61971445 and 61671158].

References

[1] B. T. Barrett, A. Bradley, and T. R. Candy, “The relationshipbetween anisometropia and amblyopia,” Progress in Retinaland Eye Research, vol. 36, pp. 120–158, 2013.

[2] D. M. Levi, D. C. Knill, and D. Bavelier, “Stereopsis and ambly-opia: a mini-review,” Vision Research, vol. 114, pp. 17–30,2015.

[3] K. Simons, “Amblyopia characterization, treatment, and pro-phylaxis,” Survey of Ophthalmology, vol. 50, no. 2, pp. 123–166, 2005.

[4] R. F. Hess, B. Mansouri, S. C. Dakin, and H. A. Allen, “Integra-tion of local motion is normal in amblyopia,” Journal of theOptical Society of America A, vol. 23, no. 5, p. 986, 2006.

[5] M. Hayhoe, B. Gillam, K. Chajka, and E. Vecellio, “The role ofbinocular vision in walking,” Visual Neuroscience, vol. 26,no. 1, pp. 73–80, 2009.

[6] E. E. Birch, “Amblyopia and binocular vision,” Progress in Ret-inal and Eye Research, vol. 33, no. 1, pp. 67–84, 2013.

[7] D. P. Spiegel, J. Li, R. F. Hess et al., “Transcranial direct currentstimulation enhances recovery of stereopsis in adults withamblyopia,” Neurotherapeutics, vol. 10, no. 4, pp. 831–839,2013.

[8] T. Li and K. Shotton, “Cochrane Review: Conventional occlu-sion versus pharmacologic penalization for amblyopia,” Evi-dence-Based Child Health: A Cochrane Review Journal, vol. 5,no. 4, pp. 1873–1909, 2010.

FP1 FPZ

AF3 AF4F7

F5 F3 F1 FZ F2 F4 F6F8

FT7 FC5 FC3 FC1 FCZ FC2 FC4 FC6 FT8

T7 C5 C3 C1 CZ C2 C4 C6 T8

TP7M1

CP5 CP3 CP1 CPZ CP2 CP4 CP6TP8

M2

P7P5

P3 P1 PZ P2 P4P6

P8

PO7PO5PO3 POZ PO4PO6

PO8O1 OZ O2

CB1 CB2

FP2

Figure 5: EEG channels showed significant correlation betweenrelative PSD change and training effect. Red colour indicates thechannels with correlation coefficient beyond 0.8, while yellowcolour indicates the channels with correlation coefficient between0.5 and 0.8.

6 Computational and Mathematical Methods in Medicine

[9] R. Bhola, R. V. Keech, P. Kutschke, W. Pfeifer, andW. E. Scott,“Recurrence of amblyopia after occlusion therapy,” Ophthal-mology, vol. 113, no. 11, pp. 2097–2100, 2006.

[10] J. Li, B. Thompson, C. S. Y. Lam et al., “The role of suppressionin amblyopia,” Investigative Opthalmology & Visual Science,vol. 52, no. 7, pp. 4169–4176, 2011.

[11] R. F. Hess, B. Mansouri, and B. Thompson, “A binocularapproach to treating amblyopia: antisuppression therapy,”Optometry and Vision Science, vol. 87, no. 9, pp. 697–704, 2010.

[12] J. Ding and D. M. Levi, “Recovery of stereopsis through per-ceptual learning in human adults with abnormal binocularvision,” Proceedings of the National Academy of Sciences ofthe United States of America, vol. 108, no. 37, pp. E733–E741, 2011.

[13] R. F. Hess, B. Thompson, and D. H. Baker, “Binocular vision inamblyopia: structure, suppression and plasticity,” Ophthalmicand Physiological Optics, vol. 34, no. 2, pp. 146–162, 2014.

[14] S. L. Li, R. M. Jost, S. E. Morale et al., “A binocular iPad treat-ment for amblyopic children,” Eye, vol. 28, no. 10, pp. 1246–1253, 2014.

[15] D. Levi, I. Vedamurthy, D. Knill, J. Ding, O.-S. Kwon, andD. Bavelier, “Recovering stereo vision by squashing virtualbugs in a VR environment,” Journal of Vision, vol. 16, no. 12,p. 201, 2016.

[16] P. Žiak, A. Holm, J. Halička, P. Mojžiš, and D. P. Piñero,“Amblyopia treatment of adults with dichoptic training usingthe virtual reality oculus rift head mounted display: preliminaryresults,” BMC Ophthalmology, vol. 17, no. 1, pp. 105–108, 2017.

[17] T. L. Ooi, Y. R. Su, D. M. Natale, and Z. J. He, “A push-pulltreatment for strengthening the ‘lazy eye’ in amblyopia,” Cur-rent Biology, vol. 23, no. 8, pp. R309–R310, 2013.

[18] J. P. Xu, Z. J. He, and T. L. Ooi, “Effectively reducing sensoryeye dominance with a push-pull perceptual learning protocol,”Current Biology, vol. 20, no. 20, pp. 1864–1868, 2010.

[19] J. P. Xu, Z. J. He, and T. L. Ooi, “Push–pull training reducesfoveal sensory eye dominance within the early visual chan-nels,” Vision Research, vol. 61, no. 61, pp. 48–59, 2012.

[20] P. J. Knox, A. J. Simmers, L. S. Gray, and M. Cleary, “Anexploratory study: prolonged periods of binocular stimulationcan provide an effective treatment for childhood amblyopia,”Investigative Opthalmology & Visual Science, vol. 53, no. 2,pp. 817–824, 2012.

[21] K. Shibata, Y. Sasaki, M. Kawato, and T.Watanabe, “Neuroim-aging evidence for 2 types of plasticity in association withvisual perceptual learning,” Cerebral Cortex, vol. 26, no. 9,pp. 3681–3689, 2016.

[22] C. S. Ho and D. E. Giaschi, “Low- and high-level motion per-ception deficits in anisometropic and strabismic amblyopia:evidence from fMRI,” Vision Research, vol. 49, no. 24,pp. 2891–2901, 2009.

[23] Z. Tan, Y. Li, D. Zang et al., “Altered regional homogeneity inepileptic patients with infantile spasm: a resting-state fMRIstudy,” Journal of X-Ray Science and Technology, vol. 24,no. 2, pp. 285–295, 2016.

[24] C. I. Yang, M. L. Yang, J. C. Huang et al., “Functional MRI ofamblyopia before and after levodopa,” Neuroscience Letters,vol. 339, no. 1, pp. 49–52, 2003.

[25] S. Gupta, S. S. Kumaran, R. Saxena, S. Gudwani, V. Menon,and P. Sharma, “BOLD fMRI and DTI in strabismicamblyopes following occlusion therapy,” International Oph-thalmology, vol. 36, no. 4, pp. 557–568, 2016.

[26] Y. Iwata, T. Handa, H. Ishikawa, N. Shoji, and K. Shimizu,“Efficacy of an amblyopia treatment program with both eyesopen: a functional near-infrared spectroscopy study,” Ameri-can Orthoptic Journal, vol. 66, no. 1, pp. 87–91, 2016.

[27] S. Deng, L. Gu, Z. L. Lu et al., “EEG Recordings during Binoc-ular Rivalry Reveals Changes of Binocular Interaction Follow-ing Perceptual Learning in Adult Amblyopia,” InvestigativeOphthalmology & Visual Science, vol. 59, 2018.

[28] A. Delorme and S. Makeig, “EEGLAB: an open source toolboxfor analysis of single-trial EEG dynamics including indepen-dent component analysis,” Journal of Neuroscience Methods,vol. 134, no. 1, pp. 9–21, 2004.

[29] P. Welch, “The use of fast Fourier transform for the estimationof power spectra: a method based on time averaging overshort, modified Periodograms,” IEEE Transactions on Audioand Electroacoustics, vol. 15, no. 2, pp. 70–73, 1967.

[30] S. Nakamura, N. Sadato, T. Oohashi, E. Nishina, Y. Fuwamoto,and Y. Yonekura, “Analysis of music-brain interaction withsimultaneous measurement of regional cerebral blood flowand electroencephalogram beta rhythm in human subjects,”Neuroscience Letters, vol. 275, no. 3, pp. 222–226, 1999.

[31] L. G. Ungerleider and J. V. Haxby, “‘What’ and ‘where’ in thehuman brain,” Current Opinion in Neurobiology, vol. 4, no. 2,pp. 157–165, 1994.

[32] B. Thompson, M. Y. Villeneuve, C. Casanova, and R. F. Hess,“Abnormal cortical processing of pattern motion in ambly-opia: evidence from fMRI,” NeuroImage, vol. 60, no. 2,pp. 1307–1315, 2012.

[33] A. J. Simmers, T. Ledgeway, B. Mansouri, C. V. Hutchinson,and R. F. Hess, “The extent of the dorsal extra-striate deficitin amblyopia,” Vision Research, vol. 46, no. 16, pp. 2571–2580, 2006.

[34] D. Leopold, “Primary visual cortex: awareness and blindsight,”Annual Review of Neuroscience, vol. 35, pp. 91–109, 2012.

[35] S. Henriksen, S. Tanabe, and B. Cumming, “Disparity process-ing in primary visual cortex,” Philosophical Transactions of theRoyal Society B: Biological Sciences, vol. 371, no. 1697, article20150255, 2016.

7Computational and Mathematical Methods in Medicine